class - threats to internal validity

class - threats to internal validity - i Homo/Heterogeneous...

Info iconThis preview shows page 1. Sign up to view the full content.

View Full Document Right Arrow Icon
1.8.3 Class Notes – Threats to Internal Validity I) Threats to Internal Validity a. Subject Self Selection – whenever sample or assignment not random (when you only conduct research on a specific group then try to generalize it) (ex – conducting experiments only on college students) i. This is a threat to internal validity because without randomizing your groups you are spreading around the idiosyncrasies of the subjects and so those certain groups could contain a confound that is affecting the DV b. Demand Characteristics – something about the experiment that biases the subjects, lets them think they know how they should act (creates confound in the patient) c. Experimenter Expectancy Effects – Experimenter sees what they want to in the data d. History – some even happens between pre and post testing that influence how they take the test (usually not an issue unless it’s a longitudinal studies) e. Maturation – changes in individual between pre/post testing f. Experimental Mortality (Attrition) – loosing subjects during experiment
Background image of page 1
This is the end of the preview. Sign up to access the rest of the document.

Unformatted text preview: i. Homo/Heterogeneous Petrition – in book notes g. Testing ( do better the second time) – pretest biases the way people do the post test (learning) i. Also, in packets of questionnaires, the order of each questionnaire could affect the way they do the one following it so a good way to solve that is to mix up the order for everyone, for each testing h. Regression to the Mean – people tend to get closer to their true score, luck (bad or good) equalizes itself. i. Evaluation Apprehension – subject behaves differently because they know they’re being observed (nervous) j. Bogus Pipeline – making the subject think that you have a direct reading of their attitudes and thoughts so they are more inclined to be honest, thinking that you can tell if they’re lying k. Good idea to run “pilot test” – test out experiment for a couple of weeks (data is worthless) to work out bugs and then debrief subjects to find out if they were let on to what the experiment was about during it...
View Full Document

{[ snackBarMessage ]}

Ask a homework question - tutors are online